Skip to main content

Thresholding

Thresholding allows you to fine-tune the behavior of policies in DynamoGuard by adjusting classification thresholds. These thresholds help balance the trade-off between metrics such as recall and false positive rate, enabling you to align policy performance with your enterprise requirements. Note: setting thresholds is optional.

Thresholding for Content Policies​

For Content Policies, thresholds are classification thresholds that determine the confidence level required to flag or block content. Adjusting these thresholds affects which content is caught as compliant or non-compliant.

  • For input content policies, prompts with scores greater than the set threshold are marked as non-compliant
  • For output content policies, responses with scores less than the set threshold are marked as non-compliant

Thresholding for Hallucination Policies​

For Hallucination Policies, thresholds are based on hallucination scores generated by the evaluation model. For hallucination policies, higher scores indicate better compliance. Thresholds indicate the following for each metric:

  1. Summarization Consistency: The probability that a response is consistent with the prompt
  2. RAG Hallucination - Input Relevance: The probability that the input is relevant to the context
  3. RAG Hallucination - Response Relevance: The probability that the model response is relevant to the context or input